Sarah Uhrig, Yoalli Rezepka Garcia, Juri Opitz, Anette Frank
TL;DR该研究旨在通过跨语言的 AMR 解析,使用单个模型来捕捉各种语言中句子的核心语义内容,发现用翻译加解析(T+P)的简单方法在几种语言上的表现比最新的全球最佳系统表现更好。
Abstract
In cross-lingualabstract meaning representation (AMR) parsing, researchers
develop models that project sentences from various languages onto their AMRs to
capture their essential semantic structures: given a sen